Wednesday, 21 March 2012

Wi-Fi Location Services for Healthcare

It's fairly common for customers to be unaware of what can be achieved with Wi-Fi location services.  I've designed and deployed several large-scale Cisco / AeroScout solutions for UK healthcare in the last few years.  So I thought I would elaborate on the technology behind these types of solution.

I refer to this technology as 'Wi-Fi location services' although you will also hear it described as Wireless Tracking, RF Identification (RFID), Real Time Location Services (RTLS), Asset Management and Context-Aware Services.

There are several vendors in this space, so I apologise if this blog doesn't cover all the options.  However I've researched this field in depth and am confident that I'm covering the front-runners (Ekahau might like to argue that point).

Architecture Overview

There are three general layers to be considered for the deployment of your Wi-Fi location service.
  • Devices - RFID tags, Wi-Fi clients.
  • Network Infrastructure - Access points, WLAN controllers, chokepoints, positioning engines and management platform.
  • Asset Management and Context-Aware Services - Clever server applications....

The diagram below is a good representation of the interaction between components.

Radio Frequency Identification (RFID) Technology

In simple terms RFID is like a barcode whereby a unique ID is used to define an object.  However, RFID is effectively two technologies and "Which type of RFID is right for me?" is usually the question I'm asked.

There are two types of RFID, 'Active' RFID is different to 'Passive' RFID:
  • Active RFID uses larger more expensive tags which transmit 802.11 beacons over several metres and make use of the WLAN infrastructure.  Devices are located by triangulation, think of Google maps with 3G phones being located using signal readings from several nearby masts.  You can get accuracy to a couple of metres if you design the WLAN correctly.  Active RFID tags require 802.11 transmitters and a battery, so at smallest they are the size of a matchbox or  a 10mm thick credit card.  Depending on the tag configuration the battery will last between 3 and 4 years if the tag is set to be fairly idle whilst out-of-motion (yes they also have motion sensors).   
  • Passive RFID uses low cost tags which are detected using short-range, low-frequency scanners.   Passive tags are battery-free and attached to the object like a sticker, they are flat and inconspicuous.  Note that passive RFID doesn't work on a WLAN, it requires it's own 'reader' infrastructure.  Think of this technology like security tags in clothes stores using security scanners in the doorway.  You can get accuracy to the nearest reader point or to an area you have define by reader 'gateways'.
Active RFID makes most sense for organisations who want an infrastructure that offers multiple services, e.g. a location system with Wi-Fi data and/or voice.  Passive RFID usually comes as a complimentary solution for objects that pass a given point, such as bed-space or a storage area.  For healthcare I found that the short-range nature of passive RFID didn't meet the requirements to track objects to within a few metres anywhere in the hospital in a cost-effective manner.  To do this with passive readers would mean readers on every doorway - too expensive.  Passive RFID is more in competition with barcode technology for things like patient and blood bag tracking (think NFC).

Network Infrastructure

So let's assume you have chosen Wi-Fi for your location tracking solution, how do you begin to make this location data work for you?
  • Tracking the position of a Wi-Fi device requires a location-ready WLAN and a positioning engine with some scaled floor plan images.
  • The WLAN access points then need to report the RSSI of the beacons received from the 802.11 device to the positioning engine.
  • You need to calibrate the positioning engine through fingerprinting exercises to get those extra few metres of accuracy.
  • The positioning engine uses the client RSSI data from a minimum of 3 surrounding access points to produce the coordinates of the object.  Which can then be displayed on the map by the management server.  See below for an example of how the triangulation works.

Key Fact! You don't need RFID technology for RTLS.  All Wi-Fi adaptors can be tracked by their radio MAC address.

Key Fact! The positioning engine isn't responsible for accuracy the site surveyor is!  High accuracy is achieved through careful WLAN design and implementation.  Map segmentation, site survey methodology, scaling of maps, access points placement, map calibration of both frequencies.  If you don't get all of this right you won't get the high accuracy you are looking for.

Asset Management

The positioning engines usually have an API which allows the Asset Management solution to draw information from the system.  The Wi-Fi MAC address of the client is used as the unique ID for the asset database.  Essentially Asset Management is achieved through the addition of meaningful information to the object, for example:
  • MAC: 'aa:bb:cc:dd:ee:ff'
  • Serial: 'S247462'
  • Tag type: 'T2000'
  • Asset ID: '0012346'
  • Category: 'Wheelchair'
  • Sub-category: 'Type 4 Child Buggy'
  • Department: 'Paediatric'
  • Asset Description: 'Sunshine buggy aged 6-10'
Before you add assets you must configure the system with zones, categories, departments and users.  This will enable asset logic.

The asset logic depends upon the features and functionality of the asset management solution. Within AeroScout MobileView you have things like location of devices by name, category, description, current location and status.  You can also generate events like zone entry/exit, status change, par-level by area, proximity, dwell, out-of-sight and low battery.  These events can in turn trigger alerts such as system notifications and messaging.  'Context-awareness' is the term used to describe the ability to be aware of the assets behaviour and state.  Some examples:

  • Location
  • Status
  • Temperature/humidity
  • Motion
  • Directional motion
The screenshot below shows context-awareness using an RFID tag with MobileView.

The Cisco/AeroScout solution is impressive and has the ability to offer huge benefits to healthcare.  I've written business case justifications and these type of systems can pay for themselves within a couple of years with even the most basic deployment model.  Primarily it saves staff search time, it allows assets to be maintained on time and avoids loan costs or fines for missing maintenance targets. 

Context-Aware-Services (CAS)

Some asset management solutions provide a basic service around the information they gather.  However if you want an advanced business system that is tailored to your requirements you will usually need to find a third-party vendor who integrates with the API of the asset management solution.

One great example I came across was Nervecentre, which is able to integrate the location of assets with portering systems, nurse call, bleep systems, etc.  This is mostly achieved through CUCM, whereby the Cisco 7925 or smartphone running Cisco Mobile has further asset info.  Using this approach the user's phone becomes their location tracker and can also be used for process management apps.

The following scenario is for a 'porter call' system using a voice and location capable WLAN with Contect-Aware Services.

A nurse raises a porter request for a Sunshine buggy from an app on her smartphone.  The system will locate the closest available sunshine buggy to the nurse and send the nearest 5 porters to the buggy a smartphone notification for a 'porter call', when a porter accepts the job he can use the app to track down the buggy.  Once he confirms he has picked up the buggy he will be given the location and phone number of the requesting nurse.  If he likes he can initiate a call to the nurse to make sure he has the right item and where it should be delivered.

The cost savings that the above scenario delivers are difficult to quantify.  There are certainly time savings for staff.  More importantly, there are cases where 'clinical incidents' occur because equipment can't be found in a timely manner - they incur large fines that could be avoided.  You also need to consider the happiness of staff and patients who can become demoralised when vital pieces of equipment can't be located for extended periods of time.  In some cases staff begin 'hoarding' equipment to ensure they can use it when they need it, this is costly behaviour that we've proven can be reversed.

In Summary

The scenario I described above is achievable and is very close to production in a leading UK hospital I have designed for.  The technology gap is currently the Workforce Management applications that integrate the   Context-Aware-Services with things like porter systems, patient systems, nursing systems and VoIP telephony systems.  There is a huge demand for mobile applications in healthcare that improve efficiency.  Nervecentre are one of the early adopters in the UK for this area and I wish them well as they are a great company.

My final thought here is that some customers will be put off by the technical complexity of the solution and the fact that it isn't cheap.  But I have evidence that the cost-benefits far outstrip the TCO... so why not invest in a futuristic and efficient 'context-aware' hospital?

Sorry! Final, final thought for customers out there... Don't try and save money on the site survey it's the most important part of the design.  Find a professional outfit with experienced engineers and expect them to want to perform their own site surveys.  This has bitten many customers before!

Wednesday, 14 March 2012

Hotspot 2.0

First generation Wi-Fi hotspots haven't been popular.  They are generally used in an emergency only.  Users dont want to make credit card payments to multiple providers as they move around.  The Pay-As-You-Go billing model never matched up with 3G which uses Mobile Operator contracts to make payment, and roaming easy.  With the introduction of 4G (LTE) offering Wi-Fi speeds the Mobile Operators had pushed PAYG Wi-Fi into a corner.

In addition to the introduction of LTE, many Mobile Operators are looking to provide 'small cell' 4G which effectively works like a Wi-Fi access point. It's an OFDM radio offering data connectivity to the Evolved Packet Core.  These "pico cell" access points are being installed to high streets and they have the benefit that interference is unlikely as each installation must be registered with the RF regulator.

Consumers are much more likely to add a 4G package to their contract than use PAYG Wi-Fi.  This change in consumer behaviour means that Wi-FI will be used where there is no mobile coverage.  Or where the user density was too great for the cellular mast to meet demand.

So, lets talk about Hotspot 2.0

What is Hotspot 2.0?
Also known as Wi-Fi Certified Passpoint, Hotspot 2.0 is a new approach to public access Wi-Fi.  The idea is for mobile devices to automatically join a Wi-Fi subscriber service whenever the user enters a Hotspot 2.0 area.  The intention is to provide better bandwidth and services-on-demand to end-users, whilst also alleviating mobile carrier infrastructure of traffic overheads.

How will it work?
Hotspot 2.0 is based on the IEEE 802.11u standard.  Which is a new set of protocols to enable cellular-like roaming.  If your device supports 802.11u and you are subscribed to a Hotspot 2.0 service you will automatically connect and roam.

Where will I get it?
I would expect early adopters will be the current Wireless ISP's (BT Openzone, the Cloud, etc) and mobile carriers (T-Mobile, Vodafone, etc).  There is already a significant footprint for WISP services in event venues, hospitality, etc.  So for example, if I am a T-Mobile subscriber and they have a partnership with BT Openzone my handset will automatically join BT Openzone HS2 locations.  The consumer contracts are held by the mobile carrier, so I would assume that the data offload would be cross charged to T-Mobile.

When will it arrive?
The word on the street is of late 2012 for Wi-Fi Alliance ratification and device support and early 2013 for usable services.
  There is certainly a demand from Wi-Fi owners (who will generate income) and end-users (who get better bandwidth and services).

How is it setup?
It's a bit too early to tell, but I would speculate that the Wi-Fi Internet Service Providers (WISP) will need a Hotspot 2.0 'integrator' to audit the WLAN for readiness.  Once everyone is happy that the WLAN meets minimum standards the integrator will configure the WLAN edge infrastructure for a single service which uses cloud-based AAA to authorise the client.  The best thing about 802.11u is that a single SSID is broadcast and carries information for multiple subscription services, so the airspace remains clean.  The client devices will be authorised using 802.1X with EAP (TLS, TTLS or SIM).  This is likely to need an app on the client side to act as a dot1x supplicant.  A key feature within 802.11u is the ability to pre-associate and test Internet availability.  This should avoid 'stranded clients' when faults exist.

Hotspot 2.0 is great in concept, a secure 'turnkey' guest service is what customers want.  They also want to revert to a third party for the 'onboarding' of user accounts.  This solution also offers revenue for the Wi-Fi owner and a free service for visitors. 'Everyones a winner' as they say.... 

The technology champions have a big task ahead to get this kind of collaborative service into production.  However, there seems to be a strong appetite and early trials are under way.  So we can assume that there are few core organisations driving progress.  Hopefully we'll see a 'Hotspot 2.0 forum' of top-level SP's who offer roaming services, with international agreements for roaming abroad.

We can also assume that there will be a minimum requirements spec for WISPs to adhere to.  From an RF design perspective the goal will be to support high-density client counts.  However this will be difficult to define, for example a football stadium is vastly different to a hotel.  So from layer 1 up there will be challenges to ensure Hotspot 2.0 services meet expectations.

I'd be interested to see how Hotspot 2.0 competes with 4G/LTE for HD environments like outdoor sports and event venues?  If anyone has compared these two technologies please feel free to comment!

Marcus Burton from CWNP has also written a great blog on this topic